Recently, dynamic random access memory ("DRAM") has been commonly used for recording and replaying speech in devices such as toys, language learning recorders, and telephone answering machines. Usually, the speech recording/playback mechanism of such devices have fixed time periods for recording and replay. The memory records background noise if the speaker enters a message shorter than the preset time period. Even a speaker-time-dependent recorder retains a portion of the previous recorded message not recorded over by a subsequent message of shorter duration.
Two conventional techniques for recording and replaying speech are:
1. The machine replays automatically after detecting a certain length of background noise. Such a circuit is illustrated in FIG. 1. In this arrangement, the input signals are recorded for a fixed period of time, as determined by the time counter, and the output is recorded for the fixed period whether or not any input signals are received, as seen in FIG. 2. Thus, the portion A of the first message is recorded over by the second message C, but the portion B of the first message is not recorded over, and is repeated during the replay of the message. This arrangement has the disadvantage that noise is played upon replay. In addition, it does not replay correctly when the speaker speaks slowly or pauses. Further, the machine replays previous messages that were not recorded over, when the second message is shorter than the first.
Thus, because the memory is not cleared before making a second recording, the previous message is replayed if a later recorded message is shorter than an earlier recorded message. For example, if the DRAM is first loaded with a message content "1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20", and a later message only has the content "A B C D", the playback will contain the message "A B C D 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20", even though the desired message is "A B C D--A B C D--A B C D-- etc."
2. In the second conventional arrangement, a central processing unit (CPU), such as a micro-processor including a DRAM, controls the start and end of the recording and replay operations (as seen for example in FIG. 3). The start and end addresses of the DRAM are saved at a determined location in the DRAM, so that it is only necessary to record during times at which speech occurs. This arrangement has the disadvantage that complicated circuits are required because of the CPU, and some waste of the DRAM space occurs.
Thus, known systems have the following disadvantages:
1. The system may replay incorrectly when the speaker speaks too slowly or when there is a pause in a sentence. PA1 2. DRAM may be required to store the start and stop addresses. PA1 3. The circuit structure may be complicated if a CPU is required. PA1 4. The time required to clear the DRAM is increased as an increasing number of DRAM circuits are required for recording.